numenta / htmpapers Goto Github PK

In the README file, you say

This repository is currently under construction and will include the source code for all scripts used on Numenta's papers.

But, AFAIK, there have been more than 2 papers already published by Numenta. I am not sure why this would take so much time to do. The last commit to this repo was in November of last year.

Hi, I'm interested in your HTM Spatial Pooler and the paper in the title.

Here's the question: the SparseNet introduced in the paper generates sparse scalar vectors. But the original HTM generates binary vectors. Can I modify the SparseNet to do the same??

I tried running train_sp.py but it didn't work.
I don't know the reason, so I want to ask for help.
Error message:
Traceback (most recent call last):
   File "/disk2/zya/htm_pro/htm-paper/frontiers/the_htm_spatial_pooler_a_neocortical_algorithm_for_online_sparse_distributed_coding/train_sp.py", line 640, in
     Metrics, expName = runSPexperiments(expConfig)
   File "/disk2/zya/htm_pro/htm-paper/frontiers/the_htm_spatial_pooler_a_neocortical_algorithm_for_online_sparse_distributed_coding/train_sp.py", line 553, in runSPexperiments
     reconstructionError(sp, testInputs, activeColumnsCurrentEpoch))
   File "/disk2/zya/env/paper-htm/lib/python2.7/site-packages/htmresearch/frameworks/sp_paper/sp_metrics.py", line 836, in reconstructionError
     numActiveColumns = int(sp._localAreaDensity * sp._numColumns) + 0.0
AttributeError: 'SpatialPooler' object has no attribute '_localAreaDensity'

See original report on HTM forum. I was able to replicate the error below:

Traceback (most recent call last):
  File "train_sp.py", line 640, in <module>
    metrics, expName = runSPexperiments(expConfig)
  File "train_sp.py", line 553, in runSPexperiments
    reconstructionError(sp, testInputs, activeColumnsCurrentEpoch))
  File "/Users/mtaylor/nta/htmresearch/htmresearch/frameworks/sp_paper/sp_metrics.py", line 836, in reconstructionError
    numActiveColumns = int(sp._localAreaDensity * sp._numColumns) + 0.0
AttributeError: 'SpatialPooler' object has no attribute '_localAreaDensity'

This looks like inappropriate intimacy code smell. The sp instance is a SWiG wrapper, and it has no _localAreaDensity property. May need to add a const accessor in the cpp bindings? cc @rcrowder

Hi

I was trying to use nab-plot but it is not getting installed and giving error nab-plot command not found.

Hi, I followed the README, but when running:

python run.py -e prototype_10

I get the following error:

I have run the code in this paper for several days, and it failed.
I guess the website in the requirements.txt is not available. https://github.com/numenta/htmpapers/blob/master/biorxiv/going_beyond_the_point_neuron/requirements.txt
Could you check it if you are not busy.Thank you very much!